ESPN Total QB Rating- Week 1 (and the end?)

[gchwood]

So to add to the argument: Not having an INT, is that worth more than 300 yds passing? That is essentially the difference between Ryan Fitzpatrick (top TQBR) and Tom Brady (3rd). I mean subjectively, having watched both games, I feel that Brady's performance was more impressive and it was in a closer game, so there was more "clutch factor".

Note: NFL QBR does indeed rate Fitzpatrick higher too.

[Oldfan]

Absolute folly.

Very good QBs rise above bad defenses and make bad defenses better. The classic case example is the New Orleans Saints, 2005-06, with and without Drew Brees.

• 2005: Aaron Brooks QB, 70.0 passer rating, 13/17 TD/INT ratio, team record 3-13, offense #31 in NFL, defense #28 in NFL (by points)
  
• 2006: Drew Brees QB: 96.2 passer rating, 26/11 TD/INT ratio, team record 10-6 (division winner), offense #5 in NFL, defense #11 in NFL (by points)
  

A very good QB improves the defense by allowing the defense to play from a lead. On Sunday, the Redskins defense went from very good to superb as the team took the lead in the second half. With a 2-TD lead, the defense did not let Manning do anything.

It's not that Eli Manning threw a INT in a losing effort. The score was tied. The INT created the losing game. If he was a great QB, he wouldn't have fallen behind. He buried himself, and couldn't dig out.

Compare Colts, with and without Peyton Manning. Their defense looks worse without Manning at QB putting pressure on the opponent.

Among QBs of a given level, the hail mary passes even out with sufficient sample sizes. The bad QBs have to throw more hail mary passes, because they fall behind more. That's on the QB more than the defense.

Great QBs are not dragged down (in the QB rating) by bad defenses. The bad defenses mainly prevent them from reaching the Super Bowl and winning it.

This is the key to understanding the Colts under Manning. They must have one of the best records in the NFL throughout Manning's career, but they appeared in only 2 Super Bowls, winning only one. Manning is good enough to lift his defense in the regular season. When he faces teams with a great QB and a great defense in the playoffs, he'll lose more often than not. Defense makes a greater difference in the playoffs, among teams with great QBs. Those clashes are relatively rare in the regular season, so a great QB can maintain a high passer rating in the regular season, even with a defense that would be bad without that QB. The QB lifts the defense against most other opponents.

Manning beat the Bears in the Super Bowl, but had only an 81.8 passer rating. He had a better passer rating against the Saints two years later (88.5), but his defense killed him that year. Brees won the Super Bowl with a 114.5 passer rating. The Saints defense was better than the Colts defense.

You are promoting the bandwagon myth of the almighty quarterback by using cherry-picked stats combined with unsupported claims.

You would like to give Drew Brees all the credit for the Saints big move from 3-13 to 10-6 in 2006 and ignore the fact that Mickey Loomis’s 2006 draft was the best in the NFC South’s history. The Saints selected tailback Reggie Bush (first round), safety Roman Harper (second), guard Jahri Evans (fourth), offensive tackle Zach Strief (seventh) and receiver Marques Colston (seventh). They also traded down twice during the course of that '06 draft to acquire veteran center Jeff Faine and veteran defensive tackle Hollis Thomas via trades.

2006 was also Sean Payton’s first season. His dynamic new offensive scheme was most certainly a huge factor in the team’s turnaround.

The Dolphins were the only other team interested in Brees as a free agent. Had Miami won the bidding, Drew Brees’s reputation would have taken a big hit. It’s likely that Drew would have elevated Cam Cameron’s 2007 team to two wins as opposed to one.

In 2010, Peyton Manning threw 11 interceptions in 3 games. Did the wheels suddenly come off of Peyton the Magnificent? No, the Colts offensive skill position players suffered a rash of injuries. Wide receivers, tight ends, and running backs, who missed actual game time during the 2010 season: Joseph Addai, Donald Brown, Dallas Clark, Austin Collie, Brody Eldridge, Pierre Garcon, Anthony Gonzalez, Mike Hart, Jacob Tamme, Blair White.

Same quarterback, same scheme, same O-line, weaker supporting cast at the skill positions -- much different results.

[Atlanta Skins Fan]

So to add to the argument: Not having an INT, is that worth more than 300 yds passing? That is essentially the difference between Ryan Fitzpatrick (top TQBR) and Tom Brady (3rd). I mean subjectively, having watched both games, I feel that Brady's performance was more impressive and it was in a closer game, so there was more "clutch factor".

Note: NFL QBR does indeed rate Fitzpatrick higher too.

Insightful question. You are getting to the core of the problem with the NFL passer rating, which the ESPN formula has managed to make worse, not better, overall.

See this discussion of how the NFL rating is calculated:

http://www.baseball-statistics.com/Greats/Century/passer-rating.htm

The NFL rating has 4 components: completion percentage, YPA, TD percentage and INT percentage. Each component has a maximum score of about 40, which is why the maximum passer rating is about 158.

Each component is graded on a curve intended to normalize performances relative to average QB performances in 1971. So, significantly beating the norms for 1971 will jack up the QB rating. This is easiest to do for completion percentage.

There is nothing in the formula to account for total yards per game or percentage of offense attributable to the QB. Those are vital omissions, leading to the situation where a QB can throw for 500 yards on 50 attempts, 3 TDs and 1 INT, with a lower passer rating than another QB with 300 yards on 30 attempts, 2 TDs and 0 INT. The second QB has higher TD percentage and lower INT percentage. The first QB is clearly more dominant, however, when the production per game is considered.

These distortions illustrate my general point that the NFL passer rating is defective mainly in its formula, but can be greatly improved using data in the box score in tandem with a better formula. ESPN has managed to retain the key defects of the NFL rating while adding a bunch of garbage and subjective overload without sufficient benefit.

McNabb's clutch rating is all you need to know to see the problem. In fact, having a clutch rating is all you need to know to see the folly.

[gchwood]

Well, I kind of understand what they are doing with the clutch rating. It isn't the measure of how well they performed in the clutch, but rather they are in situations of a close game, that makes each play "clutch." Therefore, Fitzgerald, Shaub, and Flacco (among others) would have a poor clutch rating because they were not in a close game, where each play really could matter (whether it did or not) to the results of the game.

Essentially ESPN is defining clutch differently than how most fans do, making it all that much more obscure to us.

[Atlanta Skins Fan]

Well, I kind of understand what they are doing with the clutch rating. It isn't the measure of how well they performed in the clutch, but rather they are in situations of a close game, that makes each play "clutch." Therefore, Fitzgerald, Shaub, and Flacco (among others) would have a poor clutch rating because they were not in a close game, where each play really could matter (whether it did or not) to the results of the game.

Essentially ESPN is defining clutch differently than how most fans do, making it all that much more obscure to us.

The obscurity is part of the problem. The ESPN formula is based on a false set of assumptions, which they obviously haven't bothered to test adequately.

A dominant QB is less likely to be in clutch situations, because the dominance will create larger leads in the game. So you have the situation where a QB who wheezes his way to a narrow win is considered more clutch than the QB who crushes the same opponent. Or in the case of McNabb, a QB who wheezes his way to a LOSS, with a pick-six in the mix, will be considered clutch because he threw 1 TD while the score was close.

It's madness.

[Oldfan]

...Essentially ESPN is defining clutch differently than how most fans do, making it all that much more obscure to us.

It's not all that obscure. All we need to do is read their explanation.

A 15 yard completion over the middle is not worth as much when made against a team playing a prevent defense in garbage time as a 15 yard completion up the middle that picks up a first down when the score is tied. It makes sense.

[FanboyOf91]

Total QB Rating:

Tony Romo sits to pee: 62.9

Jason Campbell: 22.8

Mark Sanchez: 17.6

Uhh....

[darrelgreenie]

That is essentially the difference between Ryan Fitzpatrick (top TQBR) and Tom Brady (3rd). I mean subjectively, having watched both games, I feel that Brady's performance was more impressive and it was in a closer game, so there was more "clutch factor".

Note: NFL QBR does indeed rate Fitzpatrick higher too.

You and the people who dislike the new QBR are probably upset because they new QBR doesn't explain why or how they arrive at the conclusion that Fitzpatrick had a higher rating then Brady.

Could it be the interception?

Or is it something else?

Yes, there was more clutch factor in Brady's game and the new QBR shows this: Clutch weigth EPA

Brady 13.2 Fitz 5.5 Clutch avg Brady 1.0 Fitz 0.7

I also think people are upset because the new QBR doesn't show what they want to believe.

And I can understand that.

Here's an example of QBR missing on a QB:Matt Hasselbeck.

I had something riding on that game so I watched most of the Titans vs. Jags.

I can tell you that Hasselbeck was ineffective for most of the game except for a fluke/lucky TD pass to Britt.

Then during a possible game winning drive he throws one of the dumbest INTs I've ever seen.

Yet, Hasselbeck was above average with a 62.1 on the new QBR.

Hasselbeck faired well in the old QBR also.

But, you don't dismiss the entire stat because of few loop holes.

---------- Post added September-13th-2011 at 09:42 AM ----------

McNabb's clutch rating is all you need to know to see the problem. In fact, having a clutch rating is all you need to know to see the folly.

Most of what you write is spurious but this is an outright fallacy.

Its actually 2 fallacies: composition and generalization.

[SirClintonPortis]

Jason Campbell once again padded his stats to a "respectable" 86.4 for QBR. But ESPN's stat states he performed like a bottom tier QB. The fact is, that the TQBR certainly reflects how he played against Denver. Oakland had multiple, multiple opportunities to blow the game wide open, and Campbell's same ole, same ole hindered them to a mere 23 points.

Mark Sanchez's is also better reflected by the TQBR. The Jets didn't win because their offense drove down the field. They were just lucky due to special teams and their defense.

---------- Post added September-13th-2011 at 09:47 AM ----------

Anyway, you can ***** about Fitzpatrick all you want, but Fitzpatrick also has the highest old QBR in this weekend as well.

[Atlanta Skins Fan]

It's not all that obscure. All we need to do is read their explanation.

A 15 yard completion over the middle is not worth as much when made against a team playing a prevent defense in garbage time as a 15 yard completion up the middle that picks up a first down when the score is tied. It makes sense.

Oldfan, I think you have a future at ESPN.

The unshakable bond you share with ESPN is believing that the burden of proof in analyzing QBs is "Hey, makes sense to me!"

This led you to overrate Beck, underrate Grossman and fail to understand why the ESPN total QB rating is worse than the NFL passer rating, for multiple reasons. (Though I agree that the NFL passer rating is defective.) Most tellingly, you are persuaded by ESPN explanations that "make sense" to you, rather than demanding proof through back-testing, which ESPN has obviously failed to do. I have previously discredited several of their theories through my own back-testing.

Your confidence in your judgment is curioius, because using the same "makes sense to me" tools, you were exposed as reaching passionately wrong conclusions about Grossman vs Beck. My confidence in Grossman was always based on provable QB theory and data, which you ignored and continue to ignore, because fundamentally you don't believe that QBs can be judged by theory and data. This will lead you to a random series of correct and incorrect conclusions, depending on whether "what makes sense" to you happens to line up with the facts.

You are a skilled debater and a charming person, but you're bringing a butter knife to a gunfight when analyzing QB / passer rating formulas. "Makes sense to me" is not the burden of proof in developing a QB / passer rating formula. You don't even believe in QB / passer ratings. FWIW, the ESPN rating is more distorted by team variables than the NFL rating, and that has been your primary objection voiced against the NFL passer rating.

[Fieldy]

For example, imagine a team down 4 with 3 seconds to go in the game. In one case, it is 3rd and goal from the 3 yard line. In the second case, it is 3rd and 10 from midfield. In the first case, it is a high pressure situation. The team can win with the right play call, the right block, or the right pass, but they can also lose the game with the wrong call, block, or pass. Whether they win or lose on that last play, people will be talking about the last play after the game. In the second case, it is unlikely a Hail Mary works, so if they don't win on that last play, people will be talking about the rest of the game, not the failure on the last play. Only if they win will people be talking about that last play; there was less pressure then. The Clutch Index reflects this by assigning a much higher value in the first case than in the second.

So they're trying to assign a "pressure value" to a play based on how much people will be talking about said play the next day at the water cooler? It's an extremely subjective way to weight the statistics. The 3rd and goal play will clearly be talked about more and is probably more stressful for the coach, but the QB needs to do the same thing he does on every 3rd and goal play - get the ball in.

It seems to me, from this explanation that the "clutch index" is very subjective.

Also, looking at the website, the score is supposed to be a percentile score? So someone with a TQBR of 90 is better than 90% of the qbs in the league? Then the top QB should be 97...because he's better than 97% of the qbs in the league (better than 32 of the 33 qbs rated). Last years top QBR was Tom Brady at 76...so 24% of the quarterbacks in the league are better than him, but he has the highest rating? None of this makes sense to me.

[darrelgreenie]

Well, I kind of understand what they are doing with the clutch rating. It isn't the measure of how well they performed in the clutch, but rather they are in situations of a close game, that makes each play "clutch." Therefore, Fitzgerald, Shaub, and Flacco (among others) would have a poor clutch rating because they were not in a close game, where each play really could matter (whether it did or not) to the results of the game.

Essentially ESPN is defining clutch differently than how most fans do, making it all that much more obscure to us.

From the new QBR key:

http://espn.go.com/nfl/story/_/id/6943920/nfl-week-1-total-qbr-leaders

Clutch Wt Avg: Average clutch weight on QB action plays. Smaller than 1 means QB was in relatively less clutch situations, greater than 1 means he was in more.

If you're looking at the above ^^^stat thinking its a measure how a QB played in clutch then you're looking at it wrong.

The better stat to look for clutch play would be:

Clutch Wt EPA: Total clutch-weighted expected points added by the QB.

-HTTR

[Sellersfan]

Jason Campbell once again padded his stats to a "respectable" 86.4 for QBR. But ESPN's stat states he performed like a bottom tier QB. The fact is, that the TQBR certainly reflects how he played against Denver. Oakland had multiple, multiple opportunities to blow the game wide open, and Campbell's same ole, same ole hindered them to a mere 23 points.

Mark Sanchez's is also better reflected by the TQBR. The Jets didn't win because their offense drove down the field. They were just lucky due to special teams and their defense.

---------- Post added September-13th-2011 at 09:47 AM ----------

Anyway, you can ***** about Fitzpatrick all you want, but Fitzpatrick also has the highest old QBR in this weekend as well.

I would suggest Raider penalties, and Heyer at RT hindered them more than Campbell.

[SirClintonPortis]

From the new QBR key:

http://espn.go.com/nfl/story/_/id/6943920/nfl-week-1-total-qbr-leaders

Clutch Wt Avg: Average clutch weight on QB action plays. Smaller than 1 means QB was in relatively less clutch situations, greater than 1 means he was in more.

If you're looking at the above stat thinking it measure how a QB played in clutch then you're looking at it wrong stat.

The better stat to look for clutch play would be: Clutch Wt EPA: Total clutch-weighted expected points added by the QB.

-HTTR

Exactly, McNabb had more "clutch" situations, but he bungled them if you looked at his negative EPA.

[darrelgreenie]

I would suggest Raider penalties, and Heyer at RT hindered them more than Campbell.

While those were hinderances; last night Campbell didn't make as many plays as he normally does or should have last night.

[Oldfan]

ESPNs QBR will likely die -- not because it isn't fundamentally better than the NFL version -- but because it isn't as popular. The people who think the QBR is a good measure of QB performance won't know or care that ESPN's version makes more sense. Besides, the fact that it's less useless doesn't make it worthy.

The "batting average" in Baseball still dominates in Baseball even though it's a poor way to grade hitters. It isn't even an average. It's a percentage, but fans have been calling it an average for a hundred years. It's a tradition. Most people love tradition, even if it's a tradition of stupidity.

The NFL's QBR isn't a totally useless stat. I use it as a quick-and-dirty measure of the team's passing game on offense and their ability to stop the passing game on defense. If

relabeled as a team stat, the NFL's QBR would not deceive so many fans.

[Atlanta Skins Fan]

Mark Sanchez's is also better reflected by the TQBR. The Jets didn't win because their offense drove down the field. They were just lucky due to special teams and their defense.

That's not being fair to Sanchez. While Romo sits to pee + blocked punt lost the game for the Cowboys, the Jets were still in the game at all because of Sanchez.

Jets rushed for only 45 yards in the game, no TDs. Sanchez passed for 335 yards and 2 TDs, and also drove for a FG. 17 points are attributable to Sanchez, 7 to the punt-block, and 3 to Romo sits to pee's INT. Sanchez is not a great QB, but his job was made harder by having no rushing game. This is a defect of both the NFL passer rating and ESPN total QB rating, which is failing to account for the QB's share of offensive production.

Here are the Sanchez scoring drives:

New York Jets at 3:16, (1st play from scrimmage 3:11)

(3:11) M.Sanchez pass deep right to S.Holmes to NYJ 30 for 17 yards (A.Ball). 1-10-NYJ 13 P4

(2:27) (No Huddle, Shotgun) L.Tomlinson up the middle to NYJ 33 for 3 yards (S.Lee; D.Ware). 1-10-NYJ 30

Two-Minute Warning

(2:00) (Shotgun) M.Sanchez pass short right to D.Keller to NYJ 44 for 11 yards (S.Lee). 2-7-NYJ 33 P5

(1:36) (No Huddle, Shotgun) M.Sanchez pass short left to L.Tomlinson ran ob at DAL 47 for 9 yards. 1-10-NYJ 44

(1:30) (Shotgun) M.Sanchez sacked at NYJ 47 for -6 yards (D.Ware). 2-1-DAL 47

(1:04) (No Huddle, Shotgun) M.Sanchez pass short right to D.Keller pushed ob at DAL 36 for 17 yards

(G.Sensabaugh).

3-7-NYJ 47 P6

(:58) (Shotgun) M.Sanchez pass short right to L.Tomlinson pushed ob at DAL 4 for 32 yards (G.Sensabaugh). 1-10-DAL 36 P7

Timeout #1 by NYJ at 00:48.

(:48) M.Sanchez pass incomplete short right to D.Mason. 1-4-DAL 4

(:44) M.Sanchez pass short right to D.Keller for 4 yards, TOUCHDOWN. 2-4-DAL 4

(8:00) M.Sanchez pass deep left to J.Cumberland ran ob at DAL 47 for 33 yards. 1-10-NYJ 20 P9

(7:27) S.Greene right tackle to DAL 43 for 4 yards (D.Ware). 1-10-DAL 47

(6:52) M.Sanchez pass short middle to D.Keller to DAL 33 for 10 yards (K.Brooking). 2-6-DAL 43 P10

(6:15) M.Sanchez pass incomplete deep left to P.Burress. 1-10-DAL 33

(6:09) (Shotgun) M.Sanchez pass short middle to S.Holmes to DAL 27 for 6 yards (F.Jones). 2-10-DAL 33

(5:30) (Shotgun) M.Sanchez pass short left to L.Tomlinson pushed ob at DAL 20 for 7 yards (M.Jenkins). 3-4-DAL 27 P11

(5:00) S.Greene right tackle to DAL 16 for 4 yards (K.Brooking). 1-10-DAL 20

(4:20) M.Sanchez pass incomplete short middle to D.Keller (K.Brooking). 2-6-DAL 16

(4:14) (Shotgun) M.Sanchez pass incomplete deep middle to D.Keller. 3-6-DAL 16

(4:06) N.Folk 34 yard field goal is GOOD, Center-T.Purdum, Holder-M.Brunell. 4-6-DAL 16

(14:46) S.Greene right tackle to NYJ 15 for -1 yards (K.Coleman; S.Lissemore). 1-10-NYJ 16

(14:04) (Shotgun) M.Sanchez pass short middle to S.Holmes to NYJ 43 for 28 yards (A.Elam). DAL-M.Jenkins was

injured during the play.

2-11-NYJ 15 P13

(13:33) (Shotgun) M.Sanchez pass short right to D.Mason to DAL 44 for 13 yards (A.Elam). 1-10-NYJ 43 P14

(13:33) (No Huddle, Shotgun) M.Sanchez pass short right to L.Tomlinson to DAL 33 for 11 yards (A.Ball). 1-10-DAL 44 P15

(12:37) (No Huddle, Shotgun) L.Tomlinson left guard to DAL 26 for 7 yards (S.Lee). 1-10-DAL 33

(12:09) (No Huddle, Shotgun) M.Sanchez pass incomplete deep left to P.Burress (B.McCann). 2-3-DAL 26

(12:03) (Shotgun) M.Sanchez pass deep left to P.Burress for 26 yards, TOUCHDOWN.

Play Challenged by Replay Assistant and Upheld.

3-3-DAL 26

Giving Sanchez a TQBR of 17.6 (and below McNabb!) is a disgrace.

[gchwood]

Giving Sanchez a TQBR of 17.6 (and below McNabb!) is a disgrace.

I agree it is a disgrace. If we are going to have subjectivity and new measures, such as clutch added in, should we not factor WINNING? Sanchez as a winning QB with 300 yds should not be below a losing QB with 39 yds. EVER.

[Conn]

It's complete horse****.

The first problem is that it's a black box. They won't document the formula.

[skins2victory]

LOL... ESPN needs to go back to the drawing board with this one!!!

[Atlanta Skins Fan]

If that was their only problem, I wouldn't have much beef with it.

[darrelgreenie]

LoL, I'm sitting at my computer having my best laugh of the day.

Classic post!

[ILikeBilly]

I am not judging the system, but McNabb's TD was early in the game and the score was close. Collins' TD was late in the game and the game was already decided. In this specific case, I can see why McNabb would be rated higher. FYI, McNabbs rating was 20.7, that still sucks!!!!!

[SirClintonPortis]

The colloquial definition of "clutch" in the context of sports involves something about players performing well under high pressure situations.

This is not definition in use for ESPN's "clutch index". "How important any play is towards changing the winning percentage of the game" is what this "clutch index" measures, not how well the player performs under high pressure situations. A quarterback in a close game is going to have far more "important" plays than one in a blowout.

---------- Post added September-13th-2011 at 04:00 PM ----------

I agree it is a disgrace. If we are going to have subjectivity and new measures, such as clutch added in, should we not factor WINNING? Sanchez as a winning QB with 300 yds should not be below a losing QB with 39 yds. EVER.

Winning is the RESULT of the game. What a QB and other players do in the game leads to the result, not the other way around. The Saints might have lost to the Packers, but Brees does not deserve to have his stat brought down because of the defense or his receiver's lack of focus.

[SkinsFTW]

Well, I kind of understand what they are doing with the clutch rating. It isn't the measure of how well they performed in the clutch, but rather they are in situations of a close game, that makes each play "clutch."

So Jason Campbell will have a great clutch rating in every game because his defense keeps it close and he will never do a thing to give the team a big lead.

Sounds like garbage to me.